Method and system for generating sorting task based on image recognition and related equipment

Abstract

The invention relates to the technical field of image recognition, in particular to a method and system for generating a sorting task based on image recognition and related equipment, and the method comprises the steps: obtaining a target image of a to-be-sorted commodity; identifying the target image, and obtaining a target commodity on the target image; analyzing the target commodity, and obtaining target information of the target commodity; generating a commodity feature vector based on the target information; comparing the feature vector with a preset feature library, and obtaining commodity information of the target commodity; order information and inventory information of the target commodity are acquired; obtaining surplus information of the target commodity based on the commodity information, the order information and the inventory information; if the surplus information does not meet the preset surplus rule, generating early warning information based on the surplus information; and if the surplus information meets the preset surplus rule, generating a sorting task based on the order information. According to the method, the sorting efficiency and accuracy can be improved.

Description

Technical Field

[0001] This application relates to the field of image recognition technology, and in particular to a method, system and related equipment for generating sorting tasks based on image recognition. Background Technology

[0002] In traditional goods sorting processes, sorting tasks are primarily generated manually. Workers need to manually record information such as the category, specifications, and quantity of goods to be sorted, then compare this information with procurement requirements to create sorting tasks. This method is not only inefficient but also prone to errors in information recording and unreasonable task allocation due to human factors, affecting the overall progress and accuracy of the sorting work. With the rapid development of the logistics industry, the requirements for sorting efficiency and accuracy are increasing. Traditional methods of manually generating sorting tasks are no longer sufficient to meet actual needs. Therefore, a new method for generating sorting tasks is urgently needed to overcome the shortcomings of existing technologies. Summary of the Invention

[0003] To help improve sorting efficiency and accuracy, this application provides a method, system, and related equipment for generating sorting tasks based on image recognition.

[0004] Firstly, this application provides a method for generating sorting tasks based on image recognition, which adopts the following technical solution: A method for generating sorting tasks based on image recognition includes: Obtain the target image of the goods to be sorted; The target image is identified, and the target product on the target image is obtained; The target product is analyzed, and its target information is obtained. Based on the target information, a product feature vector is generated; The feature vector is compared with a preset feature library to obtain the product information of the target product; Obtain the order information and inventory information of the target product; Based on the product information, the order information, and the inventory information, obtain the surplus information of the target product; If the surplus information does not meet the preset surplus rules, then an early warning message is generated based on the surplus information; If the surplus information meets the preset surplus rules, a sorting task is generated based on the order information.

[0005] By adopting the above technical solution, the target image of the goods to be sorted is first acquired. Image recognition technology is used to identify the target goods, and then the target information of the target goods is analyzed to generate a product feature vector. This feature vector is compared with a preset feature library to obtain product information, as well as the order information and inventory information of the product. Based on this information, the surplus information of the goods is calculated. If the surplus information does not meet the preset rules, an early warning is generated; if it does meet the rules, a sorting task is generated based on the order information. By automatically acquiring product information through image recognition, the traditional manual recording method is replaced, reducing human error and improving information acquisition efficiency. Combining product information, orders, and inventory, tasks or early warnings are automatically judged and generated, realizing the automation and intelligence of sorting task generation, effectively improving the overall efficiency and accuracy of sorting work, and meeting the logistics industry's demand for efficient sorting.

[0006] Optionally, the step of recognizing the target image and obtaining the target product on the target image includes: The target image is divided into several basic units, and the area of ​​each unit is defined. Obtain the target object contained in the target image; Obtain the boundary contours of the target objects respectively; Based on the boundary contour, obtain the target area of ​​the target object; Based on the target area and the unit area, obtain the target ratio; If the target ratio exceeds the first quantity threshold, the target object is marked as the target product.

[0007] Optionally, generating a sorting task based on the order information if the surplus information satisfies a preset surplus rule includes: If the surplus information satisfies the preset surplus rule, the first product category is obtained based on the product information; Based on the order information, the delivery time is obtained; Based on the first product category, obtain the first selectable sorter; Based on the first selectable sorter, obtain the corresponding sorting task to be processed; Based on the pending sorting tasks and the delivery time, the final sorting personnel are determined; Based on the final sorter and the order information, a sorting task is generated.

[0008] Optionally, obtaining the first selectable sorter based on the first product category includes: Based on the first product category, obtain the sorting history data corresponding to different sorters; Based on the first product category and the sorting history data, obtain the target number of sorting times for the product corresponding to the target sorter and the target sorting efficiency corresponding to different sorting times; If the target sorting count is greater than or equal to the second quantity threshold, then the target sorter is marked as the first selectable sorter; If the target number of sorting attempts is less than the second quantity threshold, then the highest number of sorting attempts is obtained based on the first product category and the sorting history data; Based on the target number of sorting attempts and the maximum number of sorting attempts, obtain the sorting ratio value; If the sorting ratio is greater than or equal to the third quantity threshold, then the sorting efficiency change rate is obtained based on the target sorting efficiency; If the sorting efficiency change rate meets the preset change rate requirement, then the target sorter will be selected as the first selectable sorter.

[0009] Optionally, determining the final sorter based on the pending sorting tasks and the delivery time includes: Obtain the second product category corresponding to the sorting task to be processed; Based on historical sorting data, obtain the historical sorting efficiency corresponding to the second product category; Based on the historical sorting efficiency and the sorting tasks to be processed, the cumulative sorting time corresponding to different first selectable sorters is obtained; Based on the order information and the target sorting efficiency, the estimated sorting time is obtained; Calculate the estimated completion time based on the cumulative sorting time and the estimated sorting time; Based on the estimated completion time and the delivery time, the final sorting personnel are determined.

[0010] Optionally, obtaining the final sorting personnel based on the estimated completion time and the delivery time includes: Determine whether the estimated completion time is earlier than the delivery time; If the estimated completion time is earlier than the delivery time, then the target quantity is obtained; If the target quantity is equal to 1, then the final sorting staff is determined based on the estimated completion time and the delivery time; If the target quantity is greater than 1, then obtain the corresponding second selectable sorter; Based on the sorting history data, obtain the sorting accuracy rate corresponding to different second selectable sorters; Based on the sorting accuracy rate, the final sorter is determined.

[0011] Optionally, after determining whether the estimated completion time is earlier than the delivery time, the method further includes: If there is no estimated completion time earlier than the delivery time, then based on the pending sorting tasks of the selectable sorters and the corresponding order information, the delivery time corresponding to different order information is obtained; Based on the historical sorting efficiency and the order information, obtain the individual sorting time for different selectable sorters on different pending sorting tasks; Based on the individual sorting time, the delivery time, and the estimated sorting time, obtain the timeout items; Based on the different order information, the order importance is determined; Based on the order importance, obtain the importance coefficient; A comprehensive score is obtained based on the timeout item, the corresponding importance coefficient, and the preset scoring criteria; Based on the comprehensive score, the final sorting staff will be selected.

[0012] Secondly, this application also discloses a system for generating sorting tasks based on image recognition, which adopts the following technical solution: A system for generating sorting tasks based on image recognition includes: The first acquisition module is used to acquire target images of the goods to be sorted. The second acquisition module is used to identify the target image and acquire the target product on the target image; The third acquisition module is used to analyze the target product and acquire the target information of the target product; The vector generation module is used to generate product feature vectors based on the target information; The fourth acquisition module is used to compare the feature vector with a preset feature library and acquire the product information of the target product; The fifth acquisition module is used to acquire the order information and inventory information of the target product; The sixth acquisition module is used to acquire surplus information of the target product based on the product information, the order information, and the inventory information; The early warning generation module is used to generate early warning information based on the surplus information if the surplus information does not meet the preset surplus rules. If the surplus information meets the preset surplus rules, the task generation module is used to generate sorting tasks based on the order information.

[0013] By adopting the above technical solution, the target image of the goods to be sorted is first acquired. Image recognition technology is used to identify the target goods, and then the target information of the target goods is analyzed to generate a product feature vector. This feature vector is compared with a preset feature library to obtain product information, as well as the order information and inventory information of the product. Based on this information, the surplus information of the goods is calculated. If the surplus information does not meet the preset rules, an early warning is generated; if it does meet the rules, a sorting task is generated based on the order information. By automatically acquiring product information through image recognition, the traditional manual recording method is replaced, reducing human error and improving information acquisition efficiency. Combining product information, orders, and inventory, tasks or early warnings are automatically judged and generated, realizing the automation and intelligence of sorting task generation, effectively improving the overall efficiency and accuracy of sorting work, and meeting the logistics industry's demand for efficient sorting.

[0014] Thirdly, the computer device provided in this application adopts the following technical solution: A smart terminal includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and when the processor loads the computer program, it executes the method of the first aspect.

[0015] By adopting the above technical solution, a computer program is generated based on the method of the first aspect and stored in a memory for loading and execution by a processor. Thus, a smart terminal is made based on the memory and the processor, making it convenient for users to use.

[0016] Fourthly, the computer-readable storage medium provided in this application adopts the following technical solution: A computer-readable storage medium storing a computer program that, when loaded by a processor, executes the method of the first aspect.

[0017] By adopting the above technical solution, a computer program is generated based on the method of the first aspect and stored in a computer-readable storage medium for loading and execution by a processor. The computer-readable storage medium facilitates the reading and storage of the computer program.

[0018] In summary, this application includes the following beneficial technical effects: First, target images of the goods to be sorted are acquired. Image recognition technology is used to identify the target goods, and then the target information of the target goods is analyzed to generate a product feature vector. This feature vector is compared with a preset feature library to obtain product information, as well as order information and inventory information for the product. Based on this information, the surplus information of the goods is calculated. If the surplus information does not meet the preset rules, an early warning is generated; if it does meet the rules, a sorting task is generated based on the order information. By automatically acquiring product information through image recognition, the traditional manual recording method is replaced, reducing human error and improving information acquisition efficiency. Combining product information, orders, and inventory, tasks or early warnings are automatically judged and generated, realizing the automation and intelligence of sorting task generation, effectively improving the overall efficiency and accuracy of sorting work, and meeting the logistics industry's demand for efficient sorting. Attached Figure Description

[0019] Figure 1 This is a main flowchart of a method for generating sorting tasks based on image recognition according to an embodiment of this application; Figure 2 This is a flowchart of the steps in step S102 to identify the target image and obtain the target product on the target image; Figure 3 The flowchart for step S109, "If the surplus information meets the preset surplus rules, generate a sorting task based on the order information," is as follows: Figure 4 This is the flowchart of step S303, which involves obtaining the first selectable sorter based on the first product category. Figure 5 This is the flowchart of step S305, which determines the final sorter's package based on the sorting tasks to be processed and the delivery time. Figure 6 Step S506 is a flowchart of obtaining the final sorting personnel's steps based on the estimated completion time and delivery time; Figure 7 This is the flowchart for another step after step S601, which determines whether the estimated completion time is earlier than the delivery time; Figure 8 This is a block diagram of a system for generating sorting tasks based on image recognition, according to an embodiment of this application.

[0020] Explanation of reference numerals in the attached figures: 1. First acquisition module; 2. Second acquisition module; 3. Third acquisition module; 4. Vector generation module; 5. Fourth acquisition module; 6. Fifth acquisition module; 7. Sixth acquisition module; 8. Early warning generation module; 9. Task generation module. Detailed Implementation

[0021] In the first aspect, this application discloses a method for generating sorting tasks based on image recognition.

[0022] Reference Figure 1 A method for generating sorting tasks based on image recognition, comprising steps S101 to S109: Step S101: Obtain the target image of the goods to be sorted.

[0023] Specifically, high-definition industrial cameras (1920×1080 resolution, 25fps frame rate) are deployed at the entrance of the sorting area and next to the shelves in the logistics warehouse. At the same time, sorting personnel are equipped with mobile PDA terminals with 12-megapixel cameras. When the goods to be sorted enter the sorting area via the conveyor belt or are placed on the shelves, the industrial cameras automatically capture panoramic target images of the goods stack. If the goods are scattered or located in the blind spot of the camera, the sorting personnel can manually take close-up images of individual goods through the PDA terminal to ensure that the target images can clearly present the appearance features of the goods. In this embodiment, the target image refers to the image data that contains the complete appearance features of the goods to be sorted, captured by image acquisition equipment (such as high-definition industrial cameras or mobile terminal cameras).

[0024] Step S102: Recognize the target image and obtain the target product on the target image.

[0025] Specifically, the GrabCut algorithm from the OpenCV open-source library is used to perform foreground-background segmentation on the target image, extracting all foreground regions as target objects, and analyzing the target objects to obtain the target products. In this embodiment, the target objects refer to all foreground regions in the target image except for the background, which may include products, impurities, packaging fragments, etc. The target products refer to the products that need to be sorted from the target image, excluding impurities, fragments, and other non-sorting objects.

[0026] Step S103: Analyze the target product and obtain the target information of the target product.

[0027] Specifically, the HSV color space model is used to extract the main color tone (e.g., yellow cooking oil bottle, red tomato) and color distribution ratio of the target product; the Hu moment algorithm is used to calculate seven shape parameters with translation and rotation invariance to describe the product's shape (e.g., cylindrical oil drum, cuboid cardboard box); the LBP (Local Binary Pattern) algorithm is used to extract the product's surface texture features and generate a texture histogram; if the product packaging has text labels (e.g., "5L" or "1 jin"), the Tesseract OCR engine is called to recognize the text content, integrating color, shape, texture, and text labels into target information. In this embodiment, target information refers to the key feature information extracted from the target product image, including color, shape, texture, packaging labels, etc., which is the basis for generating the product feature vector.

[0028] Step S104: Generate product feature vectors based on target information.

[0029] Specifically, the target information is quantized into a numerical vector: the color feature takes the RGB values ​​of the first 3 main colors (9 dimensions), the shape feature takes 7 Hu moment parameters, the texture feature is quantized according to 16 intervals, and the OCR-recognized text is converted into an encoding (e.g., "5L" is encoded as 001); after splicing, a product feature vector is generated, and all values ​​are normalized to the [0, 1] interval to facilitate subsequent similarity comparison; in this embodiment, the product feature vector refers to converting unstructured target information into a structured numerical vector, which is used to compare with a preset feature library to identify products.

[0030] Step S105: Compare the feature vector with the preset feature library and obtain the product information of the target product.

[0031] Specifically, the preset feature library (referring to a pre-built database containing standard feature vectors and corresponding information for various commodities) is stored in MySQL. Each record contains commodity code, name, specifications, category, and standard feature vector (the standard vector is generated by extracting and averaging a large number of commodity sample images from different angles). The similarity between the target vector and the standard vector is calculated using the Euclidean distance algorithm. The standard vector with the smallest distance and less than the threshold (set to 0.3 in this embodiment) is selected. The corresponding "commodity name (e.g., Jinjian Fragrant Rapeseed Oil 5L), specifications (5L / bottle), category (rice, flour, grains and oils), code (SP002)" is the commodity information. If the distance is greater than 0.3, it is marked as "unknown commodity" and prompts for manual confirmation.

[0032] Step S106: Obtain the order information and inventory information of the target product.

[0033] Specifically, the system connects to the Enterprise Order Management System (OMS) and Warehouse Management System (WMS) via API interfaces; by inputting a product code (e.g., SP002), the system filters orders containing that product from the OMS within the past 24 hours and extracts them as order information; the system obtains the current inventory (e.g., 5 bottles) and inventory location (3rd shelf in Zone A) of the product from the WMS as inventory information; in this embodiment, the data update delay is controlled within 10 seconds to ensure real-time performance.

[0034] Step S107: Based on product information, order information, and inventory information, obtain the surplus information of the target product.

[0035] Specifically, in this embodiment, surplus information refers to the difference between the product inventory and the order demand, which is used to determine whether the inventory meets the sorting requirements; surplus quantity = inventory quantity - order demand quantity; if the product is measured by weight (such as seafood mushrooms), the actual inventory weight (such as 3 jin) is obtained through a Bluetooth scale, and the order demand weight (such as 2 jin) is subtracted to ensure that the units are consistent.

[0036] Step S108: If the earnings information does not meet the preset earnings rules, then generate an early warning information based on the earnings information.

[0037] Specifically, the preset surplus rule is set to "surplus information ≥ 0 and ≥ safety stock (10% of order demand)". If the inventory of a product is 3 catties and the demand is 4 catties (surplus -1 catties), or the inventory is 4.3 catties and the demand is 4 catties (surplus 0.3 catties < safety stock 0.4 catties), the warning generation module will automatically generate a warning message. The warning content can be "Product SP002 has a shortage of 1 catties of inventory. It is recommended to replenish the stock before 12:00". The warning will be pushed to the inventory administrator and purchasing specialist through system pop-ups and SMS, triggering the replenishment process. In this embodiment, the preset surplus rule refers to the standard for judging whether the inventory meets the conditions for generating the sorting task, which usually includes the dual requirements of basic surplus and safety stock.

[0038] Step S109: If the surplus information meets the preset surplus rules, then a sorting task is generated based on the order information.

[0039] Specifically, if the surplus information meets the rules (e.g., 1 jin ≥ 0 and ≥ 0.4 jin), then the sorter (e.g., Zhang San) is determined according to steps S301-S306, and a sorting task order (number FT20231019001, including product information, sorter, completion deadline 16:00, delivery time 17:00, and inventory location) is generated. The task order is pushed to Zhang San's PDA terminal via the MQTT protocol, the terminal pops up an audio reminder, and the background updates the task status to "pending sorting".

[0040] The method for generating sorting tasks based on image recognition provided in this embodiment first acquires a target image of the goods to be sorted, identifies the target goods using image recognition technology, analyzes the target information of the target goods to generate a product feature vector, compares the feature vector with a preset feature library to obtain product information, and simultaneously obtains the order information and inventory information of the goods, calculates the surplus information of the goods based on this information, generates an early warning if the surplus information does not meet preset rules, and generates a sorting task based on the order information if it does meet the rules, automatically acquires product information through image recognition, replacing the traditional manual recording method, reducing human error and improving information acquisition efficiency, and automatically judges and generates tasks or early warnings by combining product information, orders and inventory, realizing the automation and intelligence of sorting task generation, effectively improving the overall efficiency and accuracy of sorting work, and meeting the logistics industry's demand for efficient sorting.

[0041] Reference Figure 2 In one embodiment of this example, step S102, which involves recognizing the target image and obtaining the target product from the target image, includes steps S201 to S206: Step S201: Divide the target image into several basic units and define the unit area.

[0042] Specifically, in this embodiment, the target image can be uniformly scaled to 1000×1000 pixels and divided into 2500 basic units (50 rows × 50 columns) at 20×20 pixels each, with a unit area of ​​20×20=400 pixels. 2 , as the basis for area calculation.

[0043] Step S202: Obtain the target objects contained in the target image.

[0044] Specifically, in this embodiment, the GrabCut algorithm is used for image segmentation: First, a rectangular box containing the foreground object is manually drawn on the target image (if it is an industrial camera that automatically captures images, the position of the rectangular box can be preset to cover the conveyor belt area). The algorithm automatically iterates 10 times to optimize the segmentation results of the foreground and background and outputs a foreground mask. Based on the mask, the foreground pixels in the image are extracted to form several independent connected regions, and each connected region is a target object.

[0045] Step S203: Obtain the boundary contours of the target objects respectively.

[0046] Specifically, Gaussian blur (convolution kernel 5×5, standard deviation 1.5) is applied to the foreground pixel region of each target object, then the gradient magnitude and direction are calculated, and finally non-edge pixels are suppressed by double thresholds (low threshold 50, high threshold 150) to obtain continuous boundary contours. In this embodiment, when extracting the boundary contours of the target object, incomplete objects with edge break lengths exceeding 10% of the total contour length need to be excluded.

[0047] Step S204: Based on the boundary contour, obtain the target area of ​​the target object.

[0048] Specifically, in this embodiment, the contourArea function is used to calculate the pixel area enclosed by each boundary contour (i.e., the pixel area of ​​the target object), and then divided by the pixel area of ​​a single basic unit (400 pixels). 2 This gives the number of basic units contained in the target object (e.g., if the pixel area of ​​a target object is 2000 pixels). 2 , Number of basic units = 2000 ÷ 400 = 5; Target area = Number of basic units × Area of ​​basic units (Here, the target area is in the unit of "number of basic units" for easy calculation of the ratio later).

[0049] Step S205: Obtain the target ratio based on the target area and the unit area.

[0050] Specifically, in this embodiment, the target ratio = target area (number of basic units) ÷ 1 (single basic unit) = target area (e.g., 5). No additional calculation is required; the value of the target area is directly used.

[0051] Step S206: If the target ratio exceeds the first quantity threshold, then mark the target object as the target product.

[0052] Specifically, the first quantity threshold refers to the area standard for determining whether a target object is a target product. It is usually set according to the image area corresponding to the smallest actual size of the product to be sorted. In this embodiment, the first quantity threshold is set to 5. If the target ratio is 6 (more than 5), it is marked as a target product; if the ratio is 3 (less than 5), it is determined to be an impurity (such as packaging fragments) and excluded from subsequent analysis.

[0053] The method for generating sorting tasks based on image recognition provided in this embodiment first divides the target image of the goods to be sorted into several basic units and defines the unit area. After identifying the target objects contained in the image, the boundary contours of each target object are obtained. The target area of ​​the target object is calculated based on the boundary contours. Then, the ratio of the target area to the unit area is used to determine whether it is a target product. If the ratio exceeds a first quantity threshold, the target object is marked as a target product. Through image unit division and area ratio analysis, the product objects with actual sorting significance can be accurately screened out, and interference elements that may exist in the image (such as small impurities, background patterns, etc.) can be eliminated, improving the accuracy of target product recognition and providing a reliable recognition basis for subsequent sorting task generation, further ensuring the accuracy of sorting work.

[0054] Reference Figure 3 In one embodiment of this example, if the surplus information in step S109 satisfies the preset surplus rule, then generating a sorting task based on the order information includes steps S301 to S306: Step S301: If the surplus information meets the preset surplus rules, obtain the first product category based on the product information.

[0055] Specifically, the "Category" field in the product information has been pre-classified according to enterprise standards (such as fresh produce, rice, flour and oil, daily necessities, and home appliances), and this field is directly extracted as the first product category; for example, if the target product is "Jinjian Fragrant Rapeseed Oil 5L", and its product information category is "rice, flour and oil", then the first product category is rice, flour and oil; in this embodiment, the first product category refers to the category to which the product to be sorted belongs (such as fresh produce, rice, flour and oil).

[0056] Step S302: Obtain the delivery time based on the order information.

[0057] Specifically, the order information includes a "required delivery time" field (e.g., "2023-10-19 17:00"). In this embodiment, the transportation time needs to be calculated in advance based on the current location and the delivery location (the transportation time varies depending on the transportation time period). For example, if the transportation time is 1 hour, the delivery time needs to be 1 hour earlier than the required delivery time (e.g., if the required delivery time is 17:00 and the transportation time is 1 hour, the delivery time is 16:00) to ensure that sufficient transportation time is reserved.

[0058] Step S303: Based on the first product category, obtain the first selectable sorter.

[0059] Specifically, in this embodiment, the first selectable sorter refers to a set of sorters who have the ability to sort the first product category and who need to meet the requirements of experience or growth potential.

[0060] Step S304: Based on the first selectable sorter, obtain the corresponding sorting task to be processed.

[0061] Specifically, in this embodiment, the "pending / in the process of sorting" tasks of the first selectable sorter are queried from the task management database. For example, Zhang San has 2 tasks (2 bottles of cooking oil and 3 bags of rice), Li Si has 1 task (1 bucket of flour), and Wang Wu has no tasks. These tasks are the pending sorting tasks. In this embodiment, the pending sorting tasks refer to the tasks that the first selectable sorter has not yet completed, which are used to assess their current workload.

[0062] Step S305: Determine the final sorter based on the pending sorting tasks and delivery time.

[0063] Specifically, the final sorter refers to the personnel selected from the first pool of sorters who can complete the current sorting task on time and with high quality.

[0064] Step S306: Generate a sorting task based on the final sorter and order information.

[0065] The image recognition-based sorting task generation method provided in this embodiment first determines the first product category based on the product information and obtains the delivery time from the order information when the surplus information meets the preset surplus rules. Then, it selects the first selectable sorters based on the first product category and obtains the sorting tasks to be processed for these sorters. After determining the final sorter by combining the sorting tasks to be processed and the delivery time, it generates sorting tasks based on the final sorter and the order information. By matching the product category with the sorter and combining the delivery time with the sorter's current workload to determine the appropriate sorter, the method achieves the accuracy and efficiency of sorting task allocation. It can ensure that the sorting tasks are matched with the sorter's capabilities and that the tasks are completed within the delivery time requirements, further improving the rationality of the sorting process and the overall efficiency.

[0066] Reference Figure 4 In one embodiment of this example, step S303, based on the first product category, obtains the first selectable sorter, including steps S401 to S407: Step S401: Based on the first product category, obtain the sorting history data corresponding to different sorters.

[0067] Specifically, the sorting history data is stored in a MongoDB database. Each data entry contains "sorter ID, product category, sorting time, sorting quantity, and sorting result (correct / incorrect)". By inputting the first product category, the system can query the historical data of all sorters sorting that product category within the past 3 months. For example, sorter Zhang San's historical data contains 20 records (soring cooking oil 10 times, rice 5 times, and flour 5 times), Li Si's contains 15 records, and Wang Wu's contains 8 records. In this embodiment, the sorting history data refers to the structured record of the sorter's past sorting work.

[0068] Step S402: Based on the first product category and sorting history data, obtain the target number of sorting operations for the target sorter and the target sorting efficiency corresponding to different sorting operations.

[0069] Specifically, in this embodiment, the target sorter refers to the sorter currently being evaluated; the target sorting count refers to the total number of times the target sorter has sorted the first category of goods in the past, reflecting the richness of their sorting experience in this category of goods; the target sorting efficiency refers to the efficiency index of the target sorter when sorting the first category of goods, usually expressed as "number of sorted items per unit time" or "time spent sorting a single item", reflecting their sorting speed.

[0070] Step S403: If the target sorting count is greater than or equal to the second quantity threshold, then mark the target sorter as the first selectable sorter.

[0071] Specifically, the second quantity threshold is the criterion for determining whether a target sorter is a first selectable sorter. In this embodiment, the second quantity threshold is set to 10 times (based on the company's historical data, sorting more than 10 times can make one proficient in the sorting process of this type of goods); for example, Zhang San (20 times ≥ 10 times) and Li Si (15 times ≥ 10 times) meet the conditions and are directly marked as first selectable sorters.

[0072] Step S404: If the target number of sorting is less than the second quantity threshold, then obtain the highest number of sorting based on the first product category and historical sorting data.

[0073] Specifically, in this embodiment, the maximum number of sorting attempts refers to the maximum number of times the first product category is sorted among all sorters.

[0074] Step S405: Obtain the sorting ratio based on the target sorting count and the maximum sorting count.

[0075] Specifically, the sorting ratio refers to the ratio of the target sorter's experience to that of the most experienced sorter, used to quantify the degree to which a novice can catch up in terms of experience; in this embodiment, the sorting ratio = target sorting count ÷ highest sorting count.

[0076] Step S406: If the sorting ratio is greater than or equal to the third quantity threshold, then obtain the sorting efficiency change rate based on the target sorting efficiency.

[0077] Specifically, in this embodiment, the third quantity threshold refers to the ratio of whether the sorter's experience has reached the training standard; the sorting efficiency change rate refers to the ratio of the target sorter's recent (e.g., the last 5 times) sorting efficiency to the earlier (e.g., the previous 5 times) sorting efficiency, reflecting the trend of skill improvement (e.g., if efficiency improves, the change rate is >1; if efficiency is stable, the change rate is ≈1).

[0078] Step S407: If the sorting efficiency change rate meets the preset change rate requirement, then the target sorter is selected as the first selectable sorter.

[0079] Specifically, in this embodiment, the preset change rate requirement refers to the efficiency standard for judging whether the sorter has growth potential. It is usually set as "sorting efficiency change rate ≥ 1" (that is, efficiency does not decrease, and preferably increases), to ensure that they can perform sorting tasks and that efficiency can be optimized in the future.

[0080] The method for generating sorting tasks based on image recognition provided in this embodiment obtains the sorting history data corresponding to each sorter based on a first product category, and extracts the target sorting frequency and corresponding sorting efficiency of the target sorter for that product category. If the target sorting frequency is not lower than a second quantity threshold, the sorter is directly marked as the first selectable sorter. If it is lower than the threshold, the sorting ratio of the target sorting frequency to the highest sorting frequency in the same category is calculated. If the ratio is not lower than a third quantity threshold, it is then determined whether the sorting efficiency change rate meets the preset requirements. If it does, the sorter is included in the first selectable sorter. By combining sorting frequency, relative frequency ratio and efficiency change trend to screen sorters from multiple dimensions, experienced sorters are given priority, while opportunities are provided for promising newcomers. This ensures that the selectable sorters have the corresponding processing capabilities, while also taking into account team training and resource optimization, thereby improving the scientificity and flexibility of sorting task allocation.

[0081] Reference Figure 5 In one embodiment of this example, step S305, based on the sorting tasks to be processed and the delivery time, determines the final sorting personnel, including steps S501 to S506: Step S501: Obtain the second product category corresponding to the sorting task to be processed.

[0082] Specifically, in this embodiment, the second product category refers to the category to which the products belong in the sorting task to be processed by the first selectable sorter.

[0083] Step S502: Based on the historical sorting data, obtain the historical sorting efficiency corresponding to the second product category.

[0084] Specifically, historical sorting efficiency refers to the average efficiency of the first selectable sorter in sorting the second product category in the past. In this embodiment, the efficiency of the target sorter in sorting each second product category is extracted from the historical sorting data. For example, Zhang San's historical efficiency in sorting "cooking oil" is 2 bottles / minute, and his historical efficiency in sorting "rice" is 1 bag / minute, which is used as a benchmark for calculating the time taken to process the task.

[0085] Step S503: Based on historical sorting efficiency and pending sorting tasks, obtain the cumulative sorting time corresponding to different first selectable sorters.

[0086] Specifically, in this embodiment, the cumulative sorting time refers to the total time required for the first selectable sorter to complete all pending sorting tasks. For example, Zhang San's pending tasks are 2, namely sorting 20 barrels of cooking oil with a capacity of 5L each and sorting 50 bags of rice with a weight of 10 jin each. His cumulative sorting time is: (20 barrels ÷ 2 bottles / minute) + (50 bags ÷ 2 bags / minute) = 35 minutes.

[0087] Step S504: Based on order information and target sorting efficiency, obtain the estimated sorting time.

[0088] Specifically, in this embodiment, the estimated sorting time refers to the time required for the first selectable sorter to complete the current new sorting task (i.e., the goods to be sorted).

[0089] Step S505: Calculate the estimated completion time based on the cumulative sorting time and the estimated sorting time.

[0090] Specifically, in this embodiment, the estimated completion time refers to the total time for the first selectable sorter to complete all pending tasks and the current new task, i.e., estimated completion time = cumulative sorting time + estimated sorting time.

[0091] Step S506: Based on the estimated completion time and delivery time, obtain the final sorting personnel.

[0092] Specifically, in this embodiment, the first optional sorter whose expected completion time is earlier than the delivery time is selected as the final sorter.

[0093] The method for generating sorting tasks based on image recognition provided in this embodiment first obtains the second product category corresponding to the sorting task to be processed, and obtains the historical sorting efficiency of this category based on historical sorting data; combining the historical sorting efficiency and the sorting task to be processed, the cumulative sorting time of different first-selectable sorters is calculated, and then the estimated sorting time is obtained according to the order information and the target sorting efficiency; the cumulative sorting time and the estimated sorting time are added together to obtain the estimated completion time, and finally the final sorter is determined based on the matching of the estimated completion time and the delivery time; by quantitatively analyzing the sorter's historical efficiency, current task volume and estimated time for new tasks, the task completion time is accurately calculated and benchmarked against the delivery time, ensuring that the finally selected sorter can complete the task efficiently within the time requirement, avoiding delivery delays caused by time estimation deviations, and further improving the rationality and timeliness of sorting task allocation.

[0094] Reference Figure 6 In one embodiment of this example, step S506, based on the estimated completion time and delivery time, obtains the final sorting personnel, including steps S601 to S606: Step S601: Determine if there is an estimated completion time earlier than the delivery time.

[0095] Step S602: If the estimated completion time is earlier than the delivery time, obtain the target quantity.

[0096] Specifically, in this embodiment, the target number refers to the number of the first optional sorters whose expected completion time is earlier than the delivery time.

[0097] Step S603: If the target quantity is equal to 1, then obtain the final sorter based on the estimated completion time and delivery time.

[0098] Specifically, in this embodiment, if the target quantity is equal to 1, it means that only one first optional sorter meets the requirements, so the first optional sorter is directly used as the final sorter.

[0099] Step S604: If the target quantity is greater than 1, then obtain the corresponding second selectable sorter.

[0100] Specifically, if the target quantity is greater than 1, it means that there are multiple first-option sorters who meet the requirements. Therefore, it is necessary to obtain second-option sorters for further analysis. In this embodiment, the second-option sorter refers to all first-option sorters whose expected completion time is earlier than the delivery time, that is, the set of sorters that simultaneously meet the "skill matching" and "time constraints".

[0101] Step S605: Based on historical sorting data, obtain the sorting accuracy rate corresponding to different second-optional sorters.

[0102] Specifically, in this embodiment, the sorting accuracy rate refers to the ratio of the number of orders correctly sorted by the second selectable sorter in the past sorting work to the total number of sorted orders. It reflects the sorting quality and is the core screening indicator after the time constraint is met.

[0103] Step S606: Based on the sorting accuracy rate, obtain the final sorter.

[0104] Specifically, in this embodiment, the second selectable sorter with the highest sorting accuracy is selected as the final sorter.

[0105] The image recognition-based sorting task generation method provided in this embodiment first determines whether there are sorters whose expected completion time is earlier than the delivery time. If so, when the number of sorters meeting the criteria is 1, they are directly identified as the final sorter. When the number is greater than 1, the sorting accuracy rate is extracted from the historical sorting data of these second-selectable sorters, and the final sorter is determined based on the accuracy rate. Under the premise of ensuring that the task can be completed on time, the selection of sorters is further optimized through accuracy rate screening. This not only ensures the timeliness of delivery but also prioritizes sorters with higher accuracy rates, reduces sorting errors, and improves sorting quality, achieving a dual guarantee of efficiency and accuracy.

[0106] Reference Figure 7 In one embodiment of this example, after determining in step S601 whether there is an estimated completion time earlier than the delivery time, steps S701 to S707 are further included: Step S701: If there is no estimated completion time earlier than the delivery time, then based on the pending sorting tasks of the selectable sorters and the corresponding order information, obtain the delivery time corresponding to different order information.

[0107] Specifically, in this embodiment, if there is no expected completion time earlier than the delivery time, it means that all first optional sorters have exceeded the delivery time of the current sorting order after completing their corresponding pending sorting tasks and the current sorting task. In this case, the delivery time of the pending sorting task corresponding to each first optional sorter is obtained.

[0108] Step S702: Based on historical sorting efficiency and order information, obtain the individual sorting time for different selectable sorters for different pending sorting tasks.

[0109] Specifically, in this embodiment, the single-item sorting time refers to the time required for the first selectable sorter to complete a single pending sorting task. It can be calculated by the number of items in a single task and the historical sorting efficiency of the corresponding items, and is used to analyze the time consumption distribution of each task.

[0110] Step S703: Obtain the timeout items based on the single-item sorting time, delivery time, and estimated sorting time.

[0111] Specifically, a timeout item refers to a task item where, after assigning a new task, the time taken for the first selectable sorter to complete a pending task or the new task exceeds the corresponding order delivery time. In this embodiment, the delivery times of the current order and all pending sorting tasks corresponding to all first selectable sorters can be arranged in chronological order. Then, based on the individual sorting times of the current order and all pending sorting tasks corresponding to all first selectable sorters, an analysis can be performed to determine how many pending sorting tasks will time out. It is worth noting that in this embodiment, adjustments can also be made based on the priority of the current order and all pending sorting tasks; furthermore, the processing order of the current order can be appropriately adjusted according to the actual situation, ensuring that the current order does not time out, thereby minimizing the number of timeout items.

[0112] Step S704: Obtain the order importance based on different order information.

[0113] Specifically, in this embodiment, order importance refers to the priority set based on order attributes (such as customer level, order amount, and whether it is an urgent order), which is usually divided into three levels: high, medium, and low (e.g., VIP customer orders are of high importance, and ordinary small orders are of low importance).

[0114] Step S705: Obtain the importance coefficient based on the order importance.

[0115] Specifically, in this embodiment, the importance coefficient refers to the quantitative weight corresponding to the importance of the order (e.g., high importance coefficient = 3, medium importance coefficient = 2, low importance coefficient = 1), which is used to calculate the degree of impact of timeout risk (important orders have a greater impact from timeout).

[0116] Step S706: Obtain the comprehensive score based on the timeout item, the corresponding importance coefficient, and the preset scoring criteria.

[0117] Specifically, the preset scoring criteria refer to the rules for calculating the comprehensive score based on the timeout items and the importance coefficient. It is usually set as "initial score of 100 points, and for each timeout item, the score of 'timeout duration × importance coefficient' is deducted". The longer the timeout duration and the more important the order, the more points are deducted. The comprehensive score refers to the final score of the first selectable sorter calculated based on the preset scoring criteria, reflecting its "minimum loss potential" in the case of task conflict. The higher the score, the smaller the impact of the timeout.

[0118] In this embodiment, the overall score can be calculated according to the following formula: Where S is the overall score; S0 is the initial score; and n is the total number of timeout items. Let be the timeout duration of the i-th timeout item; is the importance coefficient corresponding to the i-th timeout item; w is the deduction weight coefficient, which defaults to 5 and can be adjusted according to the business scenario. For example, for fresh food categories with extremely high timeliness requirements, it can be adjusted to 8; for daily necessities categories with low timeliness requirements, it can be adjusted to 3.

[0119] Step S707: Based on the overall score, obtain the final sorting personnel.

[0120] Specifically, in this embodiment, the first selectable sorter with the highest overall score is selected as the final sorter.

[0121] The image recognition-based sorting task generation method provided in this embodiment, if there are no sorters whose expected completion time is earlier than the delivery time, first obtains the order delivery time corresponding to the tasks to be processed by each select sorter, and combines it with historical sorting efficiency to obtain the individual sorting time of each task; based on the individual sorting time, delivery time, and expected sorting time of the new task, timeout items are determined, and then the order importance and corresponding importance coefficient are obtained according to the order information; finally, a comprehensive score is calculated by combining the timeout items, importance coefficient, and preset scoring criteria, and the final sorter is determined based on the score; when it is impossible to ensure that all tasks are completed on time, the comprehensive score is obtained by introducing order importance to prioritize the processing of important orders, achieve optimal resource allocation in the case of time conflicts, minimize the impact of delays in important orders, and improve the flexibility and rationality of the sorting system in dealing with complex situations.

[0122] Secondly, this application also discloses a system for generating sorting tasks based on image recognition.

[0123] Reference Figure 8 A system for generating sorting tasks based on image recognition, comprising: The first acquisition module is used to acquire target images of the goods to be sorted. The second acquisition module is used to identify the target image and acquire the target product on the target image; The third acquisition module is used to analyze the target product and acquire the target information of the target product; The vector generation module is used to generate product feature vectors based on target information; The fourth acquisition module is used to compare the feature vector with the preset feature library and obtain the product information of the target product; The fifth acquisition module is used to acquire order information and inventory information for the target product; The sixth acquisition module is used to obtain the surplus information of the target product based on product information, order information, and inventory information; The early warning generation module generates early warning information based on the earnings information if the earnings information does not meet the preset earnings rules. If the surplus information meets the preset surplus rules, the task generation module is used to generate sorting tasks based on the order information.

[0124] Thirdly, this application discloses a smart terminal, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor loads the computer program, it executes a method for generating sorting tasks based on image recognition as described in the above embodiment.

[0125] Fourthly, embodiments of this application disclose a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is loaded by a processor, it executes a method for generating sorting tasks based on image recognition as described in the above embodiments.

[0126] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for generating sorting tasks based on image recognition, characterized in that, include: Obtain the target image of the goods to be sorted; The target image is identified, and the target product on the target image is obtained; The target product is analyzed, and its target information is obtained. Based on the target information, a product feature vector is generated; The feature vector is compared with a preset feature library to obtain the product information of the target product; Obtain the order information and inventory information of the target product; Based on the product information, the order information, and the inventory information, obtain the surplus information of the target product; If the surplus information does not meet the preset surplus rules, then an early warning message is generated based on the surplus information; If the surplus information meets the preset surplus rules, a sorting task is generated based on the order information.

2. The method for generating sorting tasks based on image recognition according to claim 1, characterized in that, The step of identifying the target image and obtaining the target product on the target image includes: The target image is divided into several basic units, and the area of ​​each unit is defined. Obtain the target object contained in the target image; Obtain the boundary contours of the target objects respectively; Based on the boundary contour, obtain the target area of ​​the target object; Based on the target area and the unit area, obtain the target ratio; If the target ratio exceeds the first quantity threshold, the target object is marked as the target product.

3. The method for generating sorting tasks based on image recognition according to claim 1, characterized in that, If the surplus information satisfies a preset surplus rule, then generating a sorting task based on the order information includes: If the surplus information satisfies the preset surplus rule, the first product category is obtained based on the product information; Based on the order information, the delivery time is obtained; Based on the first product category, obtain the first selectable sorter; Based on the first selectable sorter, obtain the corresponding sorting task to be processed; Based on the pending sorting tasks and the delivery time, the final sorting personnel are determined; Based on the final sorter and the order information, a sorting task is generated.

4. The method for generating sorting tasks based on image recognition according to claim 3, characterized in that, The step of obtaining the first selectable sorter based on the first product category includes: Based on the first product category, obtain the sorting history data corresponding to different sorters; Based on the first product category and the sorting history data, obtain the target number of sorting times for the product corresponding to the target sorter and the target sorting efficiency corresponding to different sorting times; If the target sorting count is greater than or equal to the second quantity threshold, then the target sorter is marked as the first selectable sorter; If the target number of sorting attempts is less than the second quantity threshold, then the highest number of sorting attempts is obtained based on the first product category and the sorting history data; Based on the target number of sorting attempts and the maximum number of sorting attempts, obtain the sorting ratio value; If the sorting ratio is greater than or equal to the third quantity threshold, then the sorting efficiency change rate is obtained based on the target sorting efficiency; If the sorting efficiency change rate meets the preset change rate requirement, then the target sorter will be selected as the first selectable sorter.

5. The method for generating sorting tasks based on image recognition according to claim 3, characterized in that, The process of determining the final sorter based on the pending sorting tasks and the delivery time includes: Obtain the second product category corresponding to the sorting task to be processed; Based on historical sorting data, obtain the historical sorting efficiency corresponding to the second product category; Based on the historical sorting efficiency and the sorting tasks to be processed, the cumulative sorting time corresponding to different first selectable sorters is obtained; Based on the order information and the target sorting efficiency, the estimated sorting time is obtained; Calculate the estimated completion time based on the cumulative sorting time and the estimated sorting time; Based on the estimated completion time and the delivery time, the final sorting personnel are determined.

6. The method for generating sorting tasks based on image recognition according to claim 5, characterized in that, The process of obtaining the final sorting personnel based on the estimated completion time and the delivery time includes: Determine whether the estimated completion time is earlier than the delivery time; If the estimated completion time is earlier than the delivery time, then the target quantity is obtained; If the target quantity is equal to 1, then the final sorting staff is determined based on the estimated completion time and the delivery time; If the target quantity is greater than 1, then obtain the corresponding second selectable sorter; Based on the sorting history data, obtain the sorting accuracy rate corresponding to different second selectable sorters; Based on the sorting accuracy rate, the final sorter is determined.

7. The method for generating sorting tasks based on image recognition according to claim 6, characterized in that, After determining whether the estimated completion time is earlier than the delivery time, the method further includes: If there is no estimated completion time earlier than the delivery time, then based on the pending sorting tasks of the selectable sorters and the corresponding order information, the delivery time corresponding to different order information is obtained; Based on the historical sorting efficiency and the order information, obtain the individual sorting time for different selectable sorters on different pending sorting tasks; Based on the individual sorting time, the delivery time, and the estimated sorting time, obtain the timeout items; Based on the different order information, the order importance is determined; Based on the order importance, obtain the importance coefficient; A comprehensive score is obtained based on the timeout item, the corresponding importance coefficient, and the preset scoring criteria; Based on the comprehensive score, the final sorting staff will be selected.

8. A system for generating sorting tasks based on image recognition, characterized in that, include: The first acquisition module is used to acquire target images of the goods to be sorted. The second acquisition module is used to identify the target image and acquire the target product on the target image; The third acquisition module is used to analyze the target product and acquire the target information of the target product; The vector generation module is used to generate product feature vectors based on the target information; The fourth acquisition module is used to compare the feature vector with a preset feature library and acquire the product information of the target product; The fifth acquisition module is used to acquire the order information and inventory information of the target product; The sixth acquisition module is used to acquire surplus information of the target product based on the product information, the order information, and the inventory information; The early warning generation module is used to generate early warning information based on the surplus information if the surplus information does not meet the preset surplus rules. If the surplus information meets the preset surplus rules, the task generation module is used to generate sorting tasks based on the order information.

9. A smart terminal, comprising a memory and a processor, characterized in that, The memory is used to store computer programs that can run on the processor, and when the processor loads the computer program, it executes the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded by the processor, it executes the method of any one of claims 1 to 7.

Images